Sentiment Analysis KESIMPULAN DAN SARAN

Jurnal Ilmiah Komputer dan Informatika KOMPUTA 5 Edisi. .. Volume. .., Bulan 20.. ISSN : 2089-9033 tokenizing process is carried out so that the resulting separation of each word.

2.4 Implementation of Naïve Bayes Algorithm

Stage In this stage, Naïve Bayes algorithm analysis process which is important in the classification of the sources of data on its sentiment is positive or negative. In this phase there are two main processes to do the classification is the process of learning and classification process. The following is an explanation of each process. 1. Learing Process In this process naïve Bayes classifier needs to be given prior knowledge to be used as a reference in order to perform the classification of the textual data based on sentiments. In the process of teaching or learning, there are three main steps. Here are the three main steps including its explanation. a. Determination of Data Class Practice At this stage, the determination of the class of data. Determination of the class is determined with the help of users by providing an opinion on whether the search keywords included in the positive class or negative class. Here is an example of the determination of class training data are presented on Tabel 5. Tabel 5 Determining The Data Class Data Word Sentiment Class D1 food people example Positif D2 kind of cat Positif D3 how to avoid violence Positif D4 how to bully people Negatif D5 example of violence Negatif D6 good violence Negatif b. Probability At this stage, probability calculations on the data that has been determined class. Tabel 6 the calculation of the probability of each class. Tabel 6 Probability Accounting Sentime nt class Count glasses Probability D 1 D 2 D 3 D 4 D 5 D 6 Positif 3 3 4 1019 Negatif 4 3 2 919 Total 3 3 4 4 3 2 1 c. Determining The Probability of a Item Once the probability of each class is calculated, then calculated the probability of each item. Here is the formula to calculate the probability per-item. ✄ p i = Probability item f i = Frequency item f c = The total number of items based on class sentiments. The following is a calculation of the probability of each item presented on Tabel 7. Tabel 7 Count Item robability Data Sentiment Class Positive Negative good ☎ ☎ ✆ ☎ ✝ people ☎ ☎ ✆ ☎ ✝ example ☎ ☎ ✆ - kind ☎ ☎ ✆ - of ☎ ☎ ✆ ☎ ✝ cats ☎ ☎ ✆ - how ☎ ☎ ✆ ☎ ✝ do ☎ ☎ ✆ ☎ ✝ avoid ☎ ☎ ✆ - violence ☎ ☎ ✆ ✞ ✝ bully - ☎ ✝ people - ☎ ✝ 2. Classification Process In this phase will be the classification of the new data, namely as test data using naïve Bayes classifier. Here is a plot of the classification process which can be seen in Gambar 5.